Cotton gas-exchange and yield responses to organic fertilizer and composite microbial inoculant in Xinjiang: a field and exploratory omics study

Organic fertilizer and microbial inoculants may modify cotton performance under arid drip irrigation, but their interaction and molecular associations remain uncertain. We evaluated these responses in a single-site 2025 field experiment in Xinjiang with 0 or 30 t ha − 1 organic fertilizer and 0, 30, 60, or 90 kg ha − 1 composite microbial inoculant across 24 independent plots. Peak-boll-stage leaf gas exchange, harvested seed cotton yield, RNA sequencing, and untargeted LC–MS were assessed. Organic fertilizer and inoculant rate had main effects on all four gas-exchange traits; their interaction was significant only for intercellular CO 2 concentration (Ci) and transpiration rate (Tr), not net photosynthetic rate (Pn) or stomatal conductance (gs). Seed cotton yield showed both main effects without an interaction. Relative to the zero-input control (O0M0), the 30 t ha − 1 organic fertilizer plus 60 kg ha − 1 inoculant treatment (O1M2) had 58.2% higher Pn, 81.6% higher gs, 39.3% higher Ci, 58.8% higher Tr, and 24.0% higher mean seed cotton yield (6247.89 kg ha − 1 ). O1M2 had the highest numerical mean yield but was not separated from the other O1 treatments; irrigation water use efficiency shared yield’s common irrigation denominator. A reproducible functional screen identified 156 candidate differentially expressed genes, of which 147 entered an exploratory co-expression network. The blue module was negatively associated with all four gas-exchange traits after Block adjustment, but none of the 12 module–trait associations remained false-discovery-rate significant after Treatment adjustment. Across seven LC–MS comparisons, only two features met the comparison-specific false-discovery-rate criteria; exploratory integrated ordination was also Treatment-sensitive. These single-year, single-site data show treatment-associated field and transcriptional responses, with limited FDR-controlled metabolomic evidence. They do not establish a causal molecular mechanism or an application-rate recommendation. Multi-year field studies and independent functional and targeted validation are required.